Article content conversion method and device

Abstract

The present invention provides a method and apparatus for article content conversion. The method comprises: obtaining an article read request triggered by a user, the article read request carrying the user identifier of the user and the article identifier of the requested article; retrieving a user vector for the user based on the user identifier, the user vector being generated based on the user's personal information; and converting the article content based on the user vector, the article identifier, and a conversion model. Because the user vector is generated based on the user's personal information, the article converted using the user vector and the conversion model better meets the needs of different users, thereby enhancing different users' sense of identification with the article.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to an article content conversion method and device. Background Art

[0002] With the continuous development of internet technology, the amount of news, entertainment, advertising, and technological information people are exposed to daily has increased dramatically. The challenge of delivering information tailored to users' needs, such as their education level, preferences, and income, is a pressing issue.

[0003] Currently, shopping platforms can provide personalized product recommendations to different users based on their browsing and purchasing history, and can also display personalized ads in ad slots. This results in each user seeing different recommended products and ads. However, when recommending products to users through soft-text ads and news articles, the content of these articles cannot be personalized for each user. Summary of the Invention

[0004] The present invention provides an article content conversion method and device, which are used to convert article content according to individual user conditions and improve user's reading recognition.

[0005] In a first aspect, the present invention provides a method for converting article content, comprising:

[0006] Obtaining an article reading request triggered by a user, wherein the article reading request carries a user ID of the user and an article ID of the article requested to be read;

[0007] Retrieving a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user;

[0008] The content of the article is converted according to the user vector, the article identifier and the conversion model.

[0009] Optionally, before retrieving the user vector of the user according to the user identifier, the method further includes:

[0010] The user vector is generated based on the user's personal information using word embedding technology.

[0011] Optionally, converting the content of the article according to the user vector, the article identifier, and the conversion model includes:

[0012] Retrieving the article according to the article identifier;

[0013] The content of the article is converted according to the user vector and the conversion model.

[0014] Optionally, before converting the content of the article according to the user vector and the conversion model, the method further includes:

[0015] Obtaining content with tagged information in the article;

[0016] Accordingly, converting the content of the article according to the user vector and the conversion model includes:

[0017] The content with the tag information is converted according to the user vector and the conversion model.

[0018] Optionally, converting the content with the tag information according to the user vector and the conversion model includes:

[0019] Determining whether the content with the marked information is a word or a sentence;

[0020] If the content with the marked information is a word, determining the target word after the content with the marked information is converted according to the user vector and the bag-of-words model (CBOW);

[0021] Replacing the content with the marked information with the target word;

[0022] If the content with the marked information is a sentence, determining a target sentence after the content with the marked information is converted according to the user vector and the sequence conversion model Seq2Seq;

[0023] The content with the marked information is replaced with the target sentence.

[0024] Optionally, determining the target word after the content with the tag information is converted based on the user vector and the bag-of-words model (CBOW) includes:

[0025] Predicting at least one candidate word based on the words included in the context of the content with the marked information;

[0026] Obtaining a word vector corresponding to each candidate word in the at least one candidate word;

[0027] The target word is determined according to the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space.

[0028] Optionally, determining the target sentence after the content with the marked information is converted according to the user vector and a sequence conversion model Seq2Seq includes:

[0029] The sentence contained in the context of the content with the tag information is input into the sequence conversion model Seq2Seq, and the sentence output by the sequence conversion model Seq2Seq is used as the target sentence.

[0030] Optionally, the method further includes: verifying the converted article using semantic and grammatical rules.

[0031] In a second aspect, the present invention provides an article content conversion device, comprising:

[0032] An acquisition module, configured to acquire an article reading request triggered by a user, wherein the article reading request carries a user ID of the user and an article ID of the article requested to be read;

[0033] a calling module, configured to call a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user;

[0034] A conversion module is used to convert the content of the article according to the user vector, the article identifier and the conversion model.

[0035] Optionally, the above device further includes:

[0036] A generation module is used to generate the user vector based on the user's personal information using word embedding technology.

[0037] Optionally, the conversion module includes: a calling unit and a conversion unit;

[0038] The retrieving unit is configured to retrieve the article according to the article identifier;

[0039] The conversion unit converts the content of the article according to the user vector and the conversion model.

[0040] Optionally, the conversion module further includes:

[0041] An acquisition unit, configured to acquire content with marked information in the article;

[0042] The conversion unit is specifically configured to convert the content with the tag information according to the user vector and the conversion model.

[0043] Optionally, the conversion unit is specifically configured to determine whether the content with the tag information is a word or a sentence;

[0044] If the content with the marked information is a word, determining the target word after the content with the marked information is converted according to the user vector and the bag-of-words model (CBOW);

[0045] Replacing the content with the marked information with the target word;

[0046] If the content with the marked information is a sentence, determining a target sentence after the content with the marked information is converted according to the user vector and the sequence conversion model Seq2Seq;

[0047] The content with the marked information is replaced with the target sentence.

[0048] Optionally, the conversion unit is specifically configured to predict at least one candidate word based on words contained in the context of the content with the marked information if the content with the marked information is a word;

[0049] Obtaining a word vector corresponding to each candidate word in the at least one candidate word;

[0050] The target word is determined according to the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space.

[0051] Optionally, the conversion unit is specifically used to input the sentence contained in the context of the content with the marked information into the sequence conversion model Seq2Seq if the content with the marked information is a sentence, and use the sentence output by the sequence conversion model Seq2Seq as the target sentence.

[0052] Optionally, the above device further includes:

[0053] The verification module is used to verify the converted article using semantic and grammatical rules.

[0054] The article content conversion method and device provided in this embodiment, upon receiving a user-triggered article read request, first retrieves the user's user vector based on the user ID. Then, based on the user vector, article ID, and conversion model, the article content is converted to produce the converted article. Because the user vector is generated based on the user's personal information, combining this user vector with the conversion model yields an article that better meets the needs of different users, enhancing their sense of identification with the article. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is an application scenario diagram of the article content conversion method provided by the present invention;

[0056] Figure 2 Flowchart of Example 1 of the article content conversion method provided by the present invention;

[0057] Figure 3 This is a flowchart of Example 2 of the article content conversion method provided by the present invention;

[0058] Figure 4 This is a schematic diagram of an article before conversion provided by the present invention;

[0059] Figure 5 A schematic diagram of an article after conversion provided by the present invention;

[0060] Figure 6 Flowchart of Example 3 of the article content conversion method provided by the present invention;

[0061] Figure 7 A schematic diagram of the conversion process performed by the sequence conversion model Seq2Seq provided by the present invention;

[0062] Figure 8 A schematic diagram of the structure of the article content conversion device provided by the present invention;

[0063] Figure 9 This is a schematic diagram of the hardware structure of the server provided by the present invention. DETAILED DESCRIPTION

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0065] Existing technologies for product recommendations and advertising display can provide personalized recommendations based on users' browsing and purchasing history, displaying personalized ads in ad slots, and thus achieving personalized product recommendations and advertising. However, when recommending products to users through soft-text ads and news articles, the content of the articles cannot be personalized; each user sees the same article.

[0066] To address the above technical issues, the present invention provides a method and device for article content conversion. This method converts the content of an article requested by a user into content that is more tailored to the user's individual circumstances, allowing the user to identify more strongly with the article and, in turn, increasing the likelihood that the user will purchase the product recommended in the article, thereby improving the article's order conversion rate.

[0067] Figure 1 This is an application scenario diagram of the article content conversion method provided by the present invention. Figure 1 The scene diagram shown includes: a terminal, an http server cluster, and an article content conversion server cluster.

[0068] In this embodiment, a user can trigger a read request for a specific article through a terminal, which then sends the read request to an HTTP server cluster. In the prior art, upon receiving the read request, the HTTP server cluster directly returns the corresponding article to the terminal for the user to view. This results in different users viewing the same article, resulting in a low level of user identification. In contrast, in the present invention, upon receiving the read request, the HTTP server cluster sends the request to an article content conversion server cluster, which converts certain content in the article based on the user's individual needs to obtain a converted article. The converted article is then sent to the HTTP server cluster, which then returns the converted article to the user.

[0069] For example, see Figure 1 As shown, it is assumed that user 1, user 2, and user 3 all trigger a read request for article 1 through the terminal. The triggering action can be simultaneous or at different times. After the http server cluster receives the read requests for article 1 triggered by these three users, it sends the three read requests to the article content conversion server cluster. The article content conversion server cluster converts the content contained in article 1 according to the individual situation of user 1, generates article a, and sends the line of article a to the terminal through the http server cluster. The terminal further displays the article a for user 1 to view. Similarly, the article content conversion server cluster converts the content contained in article 1 according to the individual situation of user 2, generates article b, and sends the line of article b to the terminal through the http server cluster. The terminal further displays the article b for user 2 to view. Similarly, the article content conversion server cluster converts the content contained in article 1 according to the individual situation of user 3, generates article c, and sends the line of article c to the terminal through the http server cluster, and the terminal further displays the article c for user 3 to view; it can be seen that although the three users all trigger the read request of article 1, the returned articles are different. The article returned to user 1 is article a, the article returned to user 2 is article b, and the article returned to user 3 is article c. Since the individual situations of different users are different, the articles converted by the article content conversion server cluster are more in line with the needs of different users, thereby enhancing the sense of identification of different users with the article.

[0070] Figure 1 The terminal is shown as a smart phone. Optionally, the terminal may also be a desktop computer, a notebook, a personal digital assistant (PDA), a tablet computer, a game console, or other devices.

[0071] The following specific embodiments are used to describe in detail the technical solution of the present invention and how the technical solution of the present invention solves the corresponding technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0072] Figure 2 This is a flowchart of the first embodiment of the article content conversion method provided by the present invention. Figure 1 The content of the article shown is converted to a server cluster for execution.

[0073] like Figure 2 As shown, the article content conversion method provided in this embodiment includes:

[0074] S201: Obtain an article reading request triggered by a user, where the article reading request carries a user ID of the user and an article ID of the article requested to be read.

[0075] S202: Retrieve a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user.

[0076] The user vector of the user may be generated in advance and stored in a corresponding database, and when it needs to be retrieved, it can be directly retrieved from the database.

[0077] Generating a user vector can be achieved by using word embedding technology based on the user's personal information. This information may include age, gender, income level, hobbies, and / or purchasing power. The specific implementation of generating a user vector using word embedding technology can be found in the prior art and will not be further described in detail in this disclosure.

[0078] S203: Convert the content of the article according to the user vector, the article identifier, and the conversion model.

[0079] One possible way to implement this step is:

[0080] First, according to the article identifier, the article is retrieved from the corresponding database;

[0081] Then, the content of the article is converted according to the user vector and the conversion model.

[0082] Optionally, the article content may be converted based on the user vector and the conversion model as follows: first, obtaining content with tagged information in the article; and then, converting the content with tagged information based on the user vector and the conversion model.

[0083] Below Figure 1The above steps are explained by taking the processing of the read request of article 1 triggered by user 1 by the article content conversion server cluster as an example:

[0084] The read request for Article 1 triggered by User 1 includes User 1's user ID and Article 1's article ID. Upon receiving this read request, the Article Content Conversion Server Cluster retrieves User 1's user vector from the corresponding database based on User 1's user ID, and also retrieves Article 1 from the corresponding database based on Article 1's article ID. Article 1 has been previously tagged, either manually or through other means, to indicate content to be converted. After obtaining User 1's user vector and the tagged Article 1, the tagged content in Article 1 is converted directly using User 1's user vector and the corresponding conversion model, yielding Article a.

[0085] The article content conversion method provided in this embodiment, upon receiving a user-triggered article read request, first retrieves the user's user vector based on the user ID. Then, based on the user vector, article ID, and conversion model, the article content is converted to produce the converted article. Because the user vector is generated based on the user's personal information, combining this user vector with the conversion model yields an article that better meets the needs of different users, enhancing their sense of identification with the article.

[0086] The content to be converted in an article can be words or sentences. Therefore, after manually or through other methods, the content to be converted in the article is marked, and the content with the marked information in the article can be words or sentences. The following describes the conversion process in detail for each of these two cases.

[0087] The first case: the content with the marked information is a word.

[0088] Figure 3 This is a flow chart of Example 2 of the article content conversion method provided by the present invention. Figure 3 As shown, the article content conversion method provided in this embodiment includes:

[0089] S301: Obtain an article reading request triggered by a user, where the article reading request carries a user ID of the user and an article ID of the article requested to be read.

[0090] S302: Retrieve a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user.

[0091] S303: Retrieve the article according to the article identifier.

[0092] S304: Obtain content with tag information in the article.

[0093] Among them, each content with tagged information in the article corresponds to a slot position, and the subsequent conversion of the content with tagged information is to convert the content corresponding to each slot position.

[0094] The implementation process of S301-S304 can be found in the above embodiment and will not be repeated here.

[0095] S305: Determine whether the content with the mark information is a word or a sentence.

[0096] S306: If the content with the tagged information is a word, determine the target word after the content with the tagged information is converted according to the user vector and the bag-of-words model (CBOW).

[0097] S307: Replace the content with the marked information with the target word.

[0098] Among them, the possible ways to determine the target words based on the user vector and the bag-of-words model CBOW are:

[0099] Step a: predict at least one candidate word based on the words contained in the context of the content with the tag information.

[0100] Step b: Obtain the word vector corresponding to each candidate word in the at least one candidate word.

[0101] Step c: Determine the target word based on the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space.

[0102] For example, if Figure 4 As shown, the content of the article retrieved from the corresponding database through S303 is "If a car is like a lover, then a pet is like one's own child. For car owners, the status of a lover and a child is the same. When we drive our car out to play, we naturally have to take our beloved pet with us, but the pet's hair, feces and various dishonesty in the car really bother us. Don't worry, Xiaozhi will recommend some pet supplies for cars to keep you away from these troubles." Assume that the slot position in this article is Figure 4 The location of the box in the figure. Before converting the content corresponding to the slot point, the slot point must be detected first. If an unconverted slot point is detected and the content corresponding to the slot point is determined to be a word, the bag-of-words model CBOW is used to predict at least one candidate word based on the words contained in the context of the slot point. The distance between the word vector and the user vector of each candidate word is calculated in the Euclidean space, and the one with the closest distance is used as the target word. Figure 4After the above calculation is performed on the content of the slot in the article shown, the target word is "cat". Then, cat is used as the target word to replace the content corresponding to the slot, and the result is as follows Figure 5 The article content is shown in Figure 2. The bag-of-words model (CBOW) is trained in advance using the corpus.

[0103] The second case: the content with the marked information is a sentence.

[0104] Figure 6 This is a flow chart of Example 3 of the article content conversion method provided by the present invention. Figure 6 As shown, the article content conversion method provided in this embodiment includes:

[0105] S601: Obtain an article reading request triggered by a user, where the article reading request carries a user ID of the user and an article ID of the article requested to be read.

[0106] S602: Retrieve a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user.

[0107] S603: Retrieve the article according to the article identifier.

[0108] S604: Obtain content with tag information in the article.

[0109] Among them, each content with tagged information in the article corresponds to a slot position, and the subsequent conversion of the content with tagged information is to convert the content corresponding to each slot position.

[0110] The implementation process of S601-S604 can be found in the above embodiment and will not be described in detail here.

[0111] S605: Determine whether the content with the mark information is a word or a sentence.

[0112] S606: If the content with the marked information is a sentence, determine a target sentence after the content with the marked information is converted based on the user vector and the sequence conversion model Seq2Seq;

[0113] S607: Replace the content with the marked information with the target sentence.

[0114] Among them, the implementation method of determining the target sentence based on the user vector and the sequence conversion model Seq2Seq is: inputting the sentence contained in the context of the content with the marked information into the sequence conversion model Seq2Seq, and using the sentence output by the sequence conversion model Seq2Seq as the target sentence.

[0115] Specifically, the process of completing the conversion through the sequence conversion model Seq2Seq is as follows:

[0116] Step a: Parse the sentence contained in the slot point context to obtain the words contained in the context. Figure 7 As described above, x1, x2, ..., x4 represent the words contained in the slot point context, h0 represents the user vector, and h1, h2, ..., h4 represent the hidden state.

[0117] In step b, the user vector h0 and words x1, x2, ..., x4 are encoded into a context vector c as follows: h1 is calculated using h1 = f(Ux1 + Wh0 + b); h2 is calculated using h2 = f(Ux2 + Wh1 + b), and h3 and h4 are calculated similarly, where U and W are parameters; the context vector c is then calculated using h1, h2, ..., h4, for example, the last hidden state h4 is assigned to c.

[0118] In step c, the context vector c is decoded by a decoder to obtain hidden states h1', h2', and h3'. Based on h1', h2', and h3', words y1, y2, and y3 are obtained. Then, based on words y1, y2, and y3, the target sentence is obtained.

[0119] Among them, the above-mentioned sequence conversion model Seq2Seq is also trained in advance using the corpus.

[0120] After all the content with marked information in the article is converted, the converted article can be verified using semantic grammar rules.

[0121] Figure 8 This is a schematic diagram of the structure of the article content conversion device provided by the present invention. Figure 8 As shown, the article content conversion device provided in this embodiment includes:

[0122] An acquisition module 801 is configured to acquire an article reading request triggered by a user, wherein the article reading request carries a user ID of the user and an article ID of the article requested to be read;

[0123] A retrieving module 802 is configured to retrieve a user vector of the user according to the user identifier, where the user vector is generated according to the user's personal information;

[0124] The conversion module 803 is configured to convert the content of the article according to the user vector, the article identifier and the conversion model.

[0125] Optionally, the article content conversion device provided in this embodiment further includes:

[0126] The generating module 804 is configured to generate the user vector based on the user's personal information by using word embedding technology.

[0127] Optionally, the conversion module includes: a calling unit 8031 and a conversion unit 8032;

[0128] The retrieving unit 8031 is configured to retrieve the article according to the article identifier;

[0129] The conversion unit 8032 converts the content of the article according to the user vector and the conversion model.

[0130] Optionally, the conversion module further includes: an acquisition unit 8033, configured to acquire content with tag information in the article;

[0131] The conversion unit 8032 is specifically configured to convert the content with the tag information according to the user vector and the conversion model.

[0132] Optionally, the conversion unit 8032 is specifically configured to determine whether the content with the tag information is a word or a sentence;

[0133] If the content with the marked information is a word, determining the target word after the content with the marked information is converted according to the user vector and the bag-of-words model (CBOW);

[0134] Replacing the content with the marked information with the target word;

[0135] If the content with the marked information is a sentence, determining a target sentence after the content with the marked information is converted according to the user vector and the sequence conversion model Seq2Seq;

[0136] The content with the marked information is replaced with the target sentence.

[0137] Optionally, the conversion unit 8032 is specifically configured to predict at least one candidate word based on words contained in the context of the content with the marked information if the content with the marked information is a word;

[0138] Obtaining a word vector corresponding to each candidate word in the at least one candidate word;

[0139] The target word is determined according to the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space.

[0140] Optionally, the conversion unit 8032 is specifically used to input the sentence contained in the context of the content with the marked information into the sequence conversion model Seq2Seq if the content with the marked information is a sentence, and use the sentence output by the sequence conversion model Seq2Seq as the target sentence.

[0141] Optionally, the article content conversion device provided in this embodiment further includes:

[0142] The verification module 805 is used to verify the converted article using semantic and grammatical rules.

[0143] The article content conversion device provided in this embodiment can be used to execute the steps in any of the above method embodiments. Its implementation principles and technical effects are similar and will not be described in detail here.

[0144] Figure 9 Schematic diagram of the hardware structure of the server provided by the present invention. Figure 9 As shown, the server of this embodiment may include:

[0145] The memory 901 is used to store program instructions.

[0146] The processor 902 is configured to implement the method described in any of the above embodiments when the program instructions are executed. The specific implementation principles can be found in the above embodiments, which will not be described in detail in this embodiment.

[0147] The present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for converting article content described in any one of the above embodiments is implemented.

[0148] The present invention also provides a program product, which includes a computer program stored in a readable storage medium. At least one processor can read the computer program from the readable storage medium, and the at least one processor executes the computer program so that the server implements the article content conversion method described in any of the above embodiments.

[0149] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0150] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0151] In addition, the functional units in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional units.

[0152] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor (English: processor) to perform some steps of the method described in various embodiments of the present invention. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (English: Read-Only Memory, abbreviated: ROM), a random access memory (English: Random Access Memory, abbreviated: RAM), a magnetic disk, or an optical disk, etc., various media that can store program code.

[0153] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for converting article content, characterized in that: include: Obtaining an article reading request triggered by a user, wherein the article reading request carries a user ID of the user and an article ID of the article requested to be read; Retrieving a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user; Retrieving the article according to the article identifier; Obtaining content with tagged information in the article; Converting the content with the marked information according to the user vector and the conversion model, including: determining whether the content with the marked information is a word or a sentence; If the content with the marked information is a word, predicting at least one candidate word based on the words contained in the context of the content with the marked information; Obtaining a word vector corresponding to each candidate word in the at least one candidate word; Determine the target word based on the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space; Replacing the content with the marked information with the target word; If the content with the tagged information is a sentence, the sentence contained in the context of the content with the tagged information is input into the sequence conversion model Seq2Seq, and the sentence output by the sequence conversion model Seq2Seq is used as the target sentence; wherein each content with the tagged information in the article corresponds to a slot position; The conversion process of the sequence conversion model Seq2Seq is as follows: parsing the sentence contained in the slot site context to obtain the words contained in the context; Encode the user vector and the words contained in the slot context into a context vector; The context vector is decoded by the Decoder to obtain the hidden state, the target word is generated based on the hidden state, and the target sentence is generated based on the target word; The content with the marked information is replaced with the target sentence.

2. The method according to claim 1, characterized in that Before retrieving the user vector of the user according to the user identifier, the method further includes: The user vector is generated based on the user's personal information using word embedding technology.

3. The method according to any one of claims 1-2, characterized in that Also includes: The converted articles are verified using semantic grammar rules.

4. An article content conversion device, characterized in that: include: An acquisition module, configured to acquire an article reading request triggered by a user, wherein the article reading request carries a user ID of the user and an article ID of the article requested to be read; a calling module, configured to call a user vector of the user according to the user identifier, where the user vector is generated according to the personal information of the user; a conversion module, configured to convert the content of the article according to the user vector, the article identifier, and a conversion model; The conversion module includes: a calling unit and a conversion unit; The retrieving unit is configured to retrieve the article according to the article identifier; The conversion unit converts the content of the article according to the user vector and the conversion model; The conversion module also includes: An acquisition unit, configured to acquire content with marked information in the article; The conversion unit is specifically configured to convert the content with the markup information according to the user vector and the conversion model; The conversion unit is specifically configured to determine whether the content with the tag information is a word or a sentence; If the content with the marked information is a word, determining the target word after the content with the marked information is converted according to the user vector and the bag-of-words model (CBOW); Replacing the content with the marked information with the target word; If the content with the marked information is a sentence, determining a target sentence after the content with the marked information is converted according to the user vector and the sequence conversion model Seq2Seq; Replacing the content with the marked information with the target sentence; The conversion unit is specifically configured to predict at least one candidate word based on the words contained in the context of the content with the marked information if the content with the marked information is a word; Obtaining a word vector corresponding to each candidate word in the at least one candidate word; Determine the target word based on the distance between the word vector corresponding to each candidate word and the user vector in the Euclidean space; The conversion unit is specifically configured to, if the content with the tagged information is a sentence, input the sentence contained in the context of the content with the tagged information into the sequence conversion model Seq2Seq, and use the sentence output by the sequence conversion model Seq2Seq as the target sentence; wherein each content with the tagged information in the article corresponds to a slot position; The conversion process of the sequence conversion model Seq2Seq is as follows: parsing the sentence contained in the slot site context to obtain the words contained in the context; Encode the user vector and the words contained in the slot context into a context vector; The context vector is decoded by the Decoder to obtain the hidden state, the target word is generated based on the hidden state, and the target sentence is generated based on the target word; The content with the marked information is replaced with the target sentence.

5. The device according to claim 4, characterized in that Also includes: A generation module is used to generate the user vector based on the user's personal information using word embedding technology.

6. The device according to any one of claims 4-5, characterized in that Also includes: The verification module is used to verify the converted article using semantic and grammatical rules.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

8. A server, characterized in that: include: processor; as well as a memory for storing executable instructions of the processor; The processor is configured to implement the method according to any one of claims 1 to 3 by executing the executable instructions.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images